Recommender Systems

Implementation and comparision of various techniques of building recommender systems, such as:

• Collaborative Filtering
• SVD (Using Dense Matrices)
• CUR (Using Sparse Matrices)
• Latent Factor Model - SVD using Gradient Descent

Dataset

We use the Movielens 1M movie ratings dataset to train and test the various models. The datasets contain around 1 million anonymous ratings of approximately 3,900 movies made by 6,000 MovieLens users who joined MovieLens in 2000.

How to run

1. Clone this repo / click "Download as Zip" and extract the files.
2. Rename the sample_config.toml to config.toml and set the required values.
3. Ensure Python 3.7 is installed, and in your system PATH.
4. Install pipenv using pip install -U pipenv.
5. In the project folder, run pipenv install to install all python dependencies.
6. Download and extract the dataset (see Dataset section) into a new folder called dataset.
7. To run the recommender for the specific technique, run its module using pipenv run python <module_name>.py.

Contributors

• Krut Patel
• P Yedhu Tilak
• Smit Sheth
• Akhil Agrawal